Study On Handing Uncertain Data Based On Relational Databases

Although the relational model for databases provides a wide range of advantages over other data models, it lacks a comprehensive way to handle uncertain data. Uncertainty in data values is pervasive in all real-world environments and has received much attention in the literature. In some fields, such as data integration, information extraction, scientific data management, multi-media application and knowledge-based system, their data is uncertain. Several methods have been proposed some uncertain data models. However, the current models have not been able to establish an acceptable extension of the relational model. In this paper, a extension of relational model based on theory of evidence is proposed and relational structure is presented and the relational algebra extended, which can handle uncertain data effectively.This paper introduces the background of uncertain data model research and relative information first. Through to analysis deferent uncertain data types, it also gives the concept and classifies of uncertain data and analyses the challenge about relational model handling uncertain data. Next, it elaborates the ULDBs model from Stanford University and introduces ULDBs how to represent uncertain data. ULDBs'data lineage is an important concept. ULDBs are an important achievement in uncertain data model. After introducing ULDBs, based on D-S evidence theory is proposed in this paper to determine the uncertain database model DSRM and positioning the model in non-subjective uncertainty data. DSRM applies eight representations to express separately the different cases of uncertain data, put the NULL value under different circumstances as evidence of the D-S evidence theory to deal with. DSRM model introduces the concept of tuple confidence. Based on the expansion of the basic relational algebra, DSRM introduces a unified algorithm for query, and given new evidence combination rules of D-S evidence theory to update data. The main research results are as follows:Based on D-S evidence theory, we apply eight representations to express separately the different cases of uncertain data and give the definitions of uncertain database model DSRM.In the analysis of the deficiencies of NULL value in relational database, we divide NULL value into the relative and absolute NULL and remove the concept of NULL value in database models; put it into D-S evidence theory unified treatment. We expand the relational algebra to meet DSRM's need.Give a new combination rule of D-S evidence theory to handle data update.Give a unified algorithm for query.